Back to Projects

Conversational SMS Booking Bot

A GoHighLevel Conversation AI configured with intent-based handoff that books ~40% of inbound leads end-to-end without rep involvement.

2025 AI Integration Engineer AI & Bots

GHL Conversation AIClaude APISMSCalendar BookingIntent Routing

taimoorakhtar.com/projects/sms-booking-bot

Introduction

A multi-location service business was getting 80+ inbound SMS leads per week, but rep response time was 15-45 minutes on average. By the time a rep replied, 30% of leads had gone cold. The client wanted automation that could handle qualification and booking conversationally — not a rigid menu tree.

The Challenge

Most SMS bots fail at conversational nuance. They either follow scripted menus (which annoy users) or hallucinate when faced with ambiguous input (which embarrasses the brand). The challenge: build something that could handle 'I need someone to come look at my AC' AND 'my heater is making a weird clicking noise' AND 'how much for a furnace inspection' — and route each appropriately.

The Solution

Deployed GoHighLevel's Conversation AI with a custom-tuned prompt, a curated knowledge base of services and FAQs, and explicit intent-detection triggers that hand off to a human rep when confidence drops below threshold. Booking is wired directly to the GHL calendar with availability rules.

Technical Deep Dive

Custom prompt with handoff protocol. Prompt explicitly defines: 'When you cannot determine intent, when user expresses frustration, when user asks for a human, when issue is emergency-tagged — immediately respond with handoff message and tag the contact for rep review.'

Knowledge base curation. All services, pricing ranges, response-time commitments, and service area boundaries loaded into GHL's knowledge base. Bot answers only from this source — no general LLM speculation.

Calendar slot offering. Bot reads live availability from the GHL calendar via the booking API. Offers 3 slots based on the user's stated urgency. Books on confirmation, sends calendar invite and SMS reminder.

Intent-based escalation. Emergency keywords ('flooding', 'no heat', 'gas smell') trigger immediate handoff to the on-call rep with a priority tag — bot doesn't try to book these.

Conversation logs and tuning. Every conversation reviewed weekly for the first month. Edge cases identified and added to the knowledge base or prompt. Quality climbed steadily across 6 weeks.

Key Features

Full conversational booking via SMS
Intent-aware human handoff
Live calendar availability integration
Emergency keyword priority escalation
Weekly conversation review and tuning

Results & Impact

~40% of inbound SMS leads book end-to-end without rep involvement
Average response time dropped from 23 min to under 30 seconds
Rep capacity freed to focus on high-value and complex cases
Zero compliance incidents across 6 months of operation

Lessons Learned

"Conversation AI works only if you constrain it to a curated knowledge base — pure LLM is dangerous in customer-facing SMS."

"Intent-based handoff is more important than the bot's individual responses. Get the routing right, get the bot wrong gracefully."

"First 6 weeks of any conversational deployment is tuning. Plan for it, budget for it."

Related Work

Have a similar build in mind?

I'm available for engagement on GoHighLevel implementations, A2P 10DLC compliance, AI automation pipelines, and CRM migrations. Most projects start at $300–$1,200 depending on scope.

Book a 30-min call WhatsApp View More Projects